AI-Driven Accent Feedback for Pronunciation Support in Language Classrooms

People's inability to articulate their thoughts clearly is the root cause of a significant number of the difficulties they encounter while learning a second language. These issues also make it challenging for students to feel sure of themselves and speak effectively when they have to speak in front of a group. This work presents a deep learning-based framework (CNN-BiLSTM) which can apply a Convolutional Neural Network (CNN) and Bidirectional Long Short-Term Memory (BiLSTM) together to break down the speech of learners into its phonetic and temporal pieces. This framework is aimed at discovering and solving pronunciation errors that aren't extremely frequent for native speakers. Using annotated accent datasets, the model is trained and then, after each training session, it suggests adjustments in real time via a feedback mechanism. The study indicated that 93.3% of the time, the accent categorization was correct. Also, pupils' pronunciation became 34% better than when they used standard phonetic instruction. The approach could aid students who speak more than one language since it can be applied in many different ways in the classroom. This study offers a smart and scalable technique to enhance pronunciation lessons and increase student interest.

Ҳали таржима қилинмаган